LiquidAI/LFM2.5-VL-1.6B
Image-Text-to-Text
•
2B
•
Updated
•
1.23k
•
66
Maani
AI & ML interests
Novel Architectures | System Design
Recent Activity
liked
a model
1 day ago
LiquidAI/LFM2.5-VL-1.6B
liked
a model
1 day ago
LiquidAI/LFM2.5-1.2B-Instruct-GGUF
liked
a model
1 day ago
LiquidAI/LFM2.5-1.2B-Instruct
Organizations
None yet
liked
3
models
1 day ago
This model is broken.
6
#4 opened 7 days ago
by
Maani
gemma3 is absolutely trash.
#9 opened 12 days ago
by
Maani
liked
a
model
12 days ago
liked
a
model
23 days ago
upvoted
a
collection
23 days ago
liked
a
model
23 days ago
liked
a
model
28 days ago
liked
a
model
29 days ago
reacted to
sergiopaniego's
post with 🚀
29 days ago
Post
2973
NEW:
@EssentialAI
just released Rnj-1, their first 8B model.
You can easily fine-tune it with GRPO using TRL to add reasoning capabilities to a compact mode
Free Colab link: https://colab.research.google.com/github/huggingface/trl/blob/main/examples/notebooks/grpo_rnj_1_instruct.ipynb
More free TRL notebooks: https://huggingface.co/docs/trl/main/en/example_overview#notebooks
You can easily fine-tune it with GRPO using TRL to add reasoning capabilities to a compact mode
Free Colab link: https://colab.research.google.com/github/huggingface/trl/blob/main/examples/notebooks/grpo_rnj_1_instruct.ipynb
More free TRL notebooks: https://huggingface.co/docs/trl/main/en/example_overview#notebooks
After a long time, finally a capable little coder!
😎
❤️
3
7
#2 opened about 1 month ago
by
MrDevolver
reacted to
codelion's
post with 🚀
29 days ago
Post
2571
Recently, Essential AI released a new 8B base model
EssentialAI/rnj-1 they highlighted the importance of data mix for pretraning -
"In the long run, we expect our methods to automatically represent, transform, and blend data to optimize measurable abilities in pre-training. Our work on modeling data taxonomies led to new approaches for jointly clustering and mixing data distributions under data repetition penalties. Many improvements in our STEM abilities can be traced back to this. "
This resonates with the recent work we did around optimal dataset mixing for pretraining where we saw have the right mix can increase the efficiency of training -
https://huggingface.co/blog/codelion/optimal-dataset-mixing
"In the long run, we expect our methods to automatically represent, transform, and blend data to optimize measurable abilities in pre-training. Our work on modeling data taxonomies led to new approaches for jointly clustering and mixing data distributions under data repetition penalties. Many improvements in our STEM abilities can be traced back to this. "
This resonates with the recent work we did around optimal dataset mixing for pretraining where we saw have the right mix can increase the efficiency of training -
https://huggingface.co/blog/codelion/optimal-dataset-mixing
liked
a
model
29 days ago
liked
2
models
about 1 month ago
upvoted
2
collections
about 1 month ago
liked
a
dataset
about 1 month ago